Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e i n f o a b s t r a c t
Article history: Optimization of Claus sulfur recovery plants requires a reliable kinetic model to describe the reaction
Received 23 December 2015 furnace and waste heat boiler. In order to develop a suitable kinetic model for prediction of reaction
Received in revised form furnace and waste heat boiler effluents, several reaction schemes were tested and the best reaction set
5 March 2016
was selected. The kinetic parameters of the best reaction scheme were optimized and different reactor
Accepted 29 March 2016
networks were analyzed using the optimum reaction set. The modified model results were in good
Available online 31 March 2016
agreement with the experimental measured plant data both before and after waste heat boiler (The
mean absolute errors were 7.62% and 7.88% respectively). In addition, the effects of increasing ammonia
Keywords:
Claus process
and oxygen content of feed, and air to acid gas ratio on reaction furnace effluents were investigated
Reaction furnace through various case studies. An optimization study was performed to control the operating variables of
Waste heat boiler an operating sulfur recovery unit in order to obtain optimum conditions (including maximum reaction
Modeling furnace temperature), maintain H2S to SO2 ratio at constant 2 and lower COS emissions from waste heat
Optimization boiler.
Kinetic © 2016 Elsevier B.V. All rights reserved.
1. Introduction The second step involves the reaction between the un-reacted
hydrogen sulfide and sulfur dioxide over a catalytic bed at lower
Environmental regulations have become more stringent on temperatures:
sulfur dioxide emissions to atmosphere. One of the main strategies
of reducing sulfur emissions is to increase the performance of the 2H2S þ SO2 4 1.5S2 þ 2H2O (3)
sulfur recovery units (SRU). The modified Claus process is one of the
major processes which converts toxic hydrogen sulfide to Conversion of H2S to SO2 in the modified Claus process is con-
elemental sulfur from acid gas during sour natural gas processing ducted in the reaction furnace (RF) which is a refractory lined cy-
and refinery upgrading (Dowling et al., 1995). The modified Claus lindrical vessel. Acid gas stream at a pressure typically ranging from
process in which hydrogen sulfide is converted to sulfur can be 130 to 180 kPa is fed to the reaction furnace burner along with an
depicted by the following simple overall reaction: appropriate quantity of air to oxidize feed contaminates and results
in a 2:1 ratio of H2S:SO2 in the reactor effluent. The combustion of
H2S þ 0.5O2 / 0.5S2 þ H2O (1) acid gases in the RF is performed at temperature ranges of
975e1300 C and gas residence times 0.5e2.0 s (Monnery et al.,
The sulfur recovery in the modified Claus process is obtained via 1993).
the two reaction steps including thermal and catalytic steps. The Since the combustion process in the RF provides substantial
first step is a thermal step where one third of hydrogen sulfide sulfur recovery and affects the entire plant emissions, any modifi-
(H2S) is partially oxidized and sulfur dioxide (SO2) is produced at cation in the RF affects sulfur conversion and emissions from Claus
very high temperatures. plant (Monnery et al., 1993). Therefore any improvement in the
reaction furnace has direct impact on reduction of sulfur emissions
H2S þ 1.5O2 / SO2 þ H2O (2) to the atmosphere.
The waste heat boiler (WHB) which is a following equipment
after the RF cools the combusted products of the RF from
926e1300 C to about 230e370 C in one or two tube passes pro-
* Corresponding author.
ducing steam on the shell side (Monnery et al., 1993).
E-mail address: ganjih@ripi.ir (H. Ganji). Optimization of the modified Claus process for further sulfur
http://dx.doi.org/10.1016/j.jngse.2016.03.086
1875-5100/© 2016 Elsevier B.V. All rights reserved.
748 S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757
recovery and lower emissions requires knowledge of reaction ki- carried out an integrated process-energy optimization at the total
netics of numerous reactions occurring in the RF and WHB (Karan, plant scale (2014). Javanmardi and Fatemi (2015) modeled a sulfur
1998). However, the available models do not consider the kinetic recovery unit based on straight through sour natural gas with two
limitations of various side reactions in the RF and WHB. These different approaches of equilibrium and kinetic methods. They
models can not properly predict CO, H2, COS and CS2 concentrations concluded that the kinetic model has more accurate results than
(Karan et al., 1999a). equilibrium model.
Although several researchers have investigated the modeling of Jones et al. (2012) identified a reaction set including indepen-
the Claus plant or specific unit operations of the Claus process dent reactions and used a four stage method to describe the
(Bennett and Meisen, 1973; Jones et al., 2012; Karan et al., 1999b; product distribution of measured WHB effluents. After modeling of
Khudenko et al., 1993; Manenti et al., 2012a, 2013b; Monnery Claus plant using the developed reaction sets, two optimization
et al., 1993; Nasato, 1993; Nasato et al., 1994; ZareNezhad and studies were carried out. These studies showed that there exists an
Hosseinpour, 2008), but most of these modeling works have used optimal steam pressure and H2S/SO2 ratio that balances hydrogen
restricted equilibrium models to estimate product distribution yield, oxygen demand, and power generation. Jones et al. (2012) did
from the Claus reaction furnace (Bennett and Meisen, 1973; not compare calculated RF outlet with measured experimental data
Khudenko et al., 1993; Monnery et al., 1993; ZareNezhad and at this equipment and their comparison was limited to the waste
Hosseinpour, 2008). Bennett and Meisen (1973) considered 36 heat boiler effluents.
chemical species including many nitrogen compounds and radicals, In this study, reaction furnace and waste heat boiler were
and reported equilibrium calculations in H2S-Air system up to a modeled both separately and incorporated via two different reac-
temperature 2000 K. tion schemes consisting same rate expressions. For this purpose,
Khudenko et al. (1993) have computed the equilibria of Claus different reaction schemes were examined as to their efficiencies in
reaction furnace under oxygen enrichment conditions employing predicting reaction furnace emissions; the reaction scheme with
Gibbs energy minimization method and concluded that oxygen- fewer errors was then chosen. To reach a higher accuracy, different
based Claus process can be used for reducing the equipment size. reactor networks were tested and fitting kinetic parameter
Monnery et al. (1993) applied the various methods such as the methods were employed.
Gibbs energy minimization method, Fisher monograph and west- The main difference between the present study and that of Jones
ern research correlations in order to estimate the reaction furnace et al. (2012) is the reaction schemes applied in modeling the re-
effluents and concluded that results of these various methods are action furnace and the waste heat boiler. Jones et al. (2012)
not in agreement with plant data taken both after and before the considered a reaction scheme, added other reactions and fitted
waste heat boiler. Selim et al. (2008) investigated critical roles of their kinetic parameters using the waste heat boiler effluents in
gases accompanied with the H2S in the sulfur recovery process on order to reach a higher accuracy while in the present research, the
determining the optimum operating temperature to obtain the kinetic parameters of the chosen reaction scheme were fitted in
maximum sulfur recovery. ZareNezhad and Hosseinpour (2008) agreement with the reaction furnace and waste heat boiler efflu-
applied chemical equilibrium calculation in order to investigate ents. After modeling the reaction furnace and waste heat boiler
application of different alternatives for increasing the RF temper- with sufficient errors, case studies were conducted so as to inves-
ature of Claus units. Some researchers have investigated the effects tigate the influence of ammonia and oxygen increments in acid gas
of acid gas feed composition on Claus process experimentally feed and air to acid gas molar flow ratio on product distribution and
(Ibrahim et al., 2013, 2014; Selim et al., 2011, 2012a, 2012b, 2013, the temperature of the reaction furnace. In addition, one optimi-
2014). zation study was done to determine the values of reaction furnace
Nasato et al. (1994) developed a rigorous one dimensional feed temperature, steam pressure and the air to acid gas molar flow
model of the WHB tube with a reaction set composed of H2S ratio in order to further optimize the reaction furnace temperature,
decomposition and COS formation reactions. In the next work by carbonyl sulfide emission, sulfur production and H2S to SO2 ratio.
Karan et al. (1994), radiation heat transfer was considered in the
reaction quenching of the waste heat boiler. It was the main dif- 2. Plant data
ference between two studies (Karan et al., 1994; Nasato et al., 1994)
and enhanced performance of the model in prediction of gas ef- In this work, the parameters and data were obtained from the
fluents. Karan et al. (1999b) used of new kinetic rate expression for Ultramar Refinery Field (Sames et al., 1991). Sames et al. (1991)
COS decomposition to model WHB whose absolute average de- measured the plant data by sampling directly from the rear
viations for COS concentrations was in 5e13%.In addition, only the chamber of the reaction furnace using high quench rates. Tests
WHB was modeled and the feed composition and other properties were done at various oxygen enrichment levels, feed compositions
were obtained from the Sames et al. (1991). and residence times. Such data were the only available measured
The other works in case of the kinetic modeling are based on a data at the reaction furnace outlet. The specific objective of their
detailed kinetic scheme for prediction of the Claus reaction furnace study was to measure the quantity of species at the reaction furnace
effluents (Manenti et al., 2013a, 2013b, 2012b; Pierucci et al., 2004; and then to determine the degree of reaction occurring in the waste
Scho€ neberger et al., 2007). Pierucci et al. (2004) modeled the heat boiler. In this regard, measured quantities of compounds at the
thermal reactor with a detailed kinetic scheme based on a radical waste heat boiler outlet were also reported. Details of feed stream
approach including 130 species and more than 1500 elementary compositions, test conditions, product distribution both prior to
reactions. and following the waste heat boiler are given in Sames et al. (1991).
Manenti et al. (2013b) used a kinetic model consist of 146 spe-
cies and 2412 reactions to model sulfur recovery unit thermal stage 3. Model
including the RF and the WHB. Their model was validated on the
industrial data of sulfur recovery units operating in the Nanjing and The reaction furnace is composed of two zones, namely the
Mumbai plants. In an another research, Manenti et al. (2013a) used flame or oxygen rich zone and post flame or anoxic zone. The flame
a kinetic model with 2400 reactions and 140 species in a proper zone is an oxygen consuming zone and its temperature exceeds
reactor network to optimize elemental sulfur recovery and steam 2000 C. Whereas residence time of the flame zone is very short
generation. Manenti et al. (2014) utilized this kinetic scheme and and in order of milliseconds, residence time in the oxygen deficient
S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757 749
post flame zone may range from 0.5 to 2.0 s depending on the flow following ordinary differential equations including mass and en-
rate and temperature (Chin et al., 2001). Thus, it seems that ergy balance equations and actual mean residence time should be
different reactor networks such as a combination of CSTR and PFR solved.
reactors in series and parallel modes would model these two zones
of RF. The WHB is made of one or two tube passes with anoxic zone dFj X
nature where hot gases from RF are cooled down to 250e300 C.
¼A yij rij j ¼ 1; …; nc i ¼ 1; …; nR (5)
dz i
Supposing negligible radial concentration and temperature gradi-
ents, one dimensional plug flow assumption can be applied in the
dT X X
RF and WHB modeling. Due to the completely different natures of ¼A hj yij rij (6)
these equipments, i.e. RF and WHB, different reaction schemes dz j i
were made use of in their modeling.
The literature reports a wide number of possible reactions for 0 11
the RF, most of which are oxidation reactions (Monnery et al., 1993). dtR 1 @
X
Table 1 shows the candidate reactions list reported in the literature ¼ ðAPÞðRTÞ Fj A (7)
dz j
for complete description of the RF. As seen in Table 1, for certain
reactions, more than one reaction rate expression is given because Above equations can be rearranged and the mass and energy
of such different experimental conditions as temperature, pressure balance equations can be written based on residence time as
and reactor type. In the present work, apart from the reactor type, follow:
we specified un-catalytic gas phase rate expressions at high tem-
peratures. It is possible to select the best kinetic expression by
P
RT Fj
examining it in a suitable model. Accordingly, about 168,000 re- dFj j X
¼ yij rij j ¼ 1; …; nc i ¼ 1; …; nR (8)
action schemes were considered and in certain cases presence or dtR P i
absence of a specific reaction was further investigated. The proce-
dure of selecting reaction schemes is as follows. P
Based on Table 1, eight rate expressions have been reported for RT Fj
dT j X X
H2S decomposition reaction. Therefore we have nine alternatives ¼ hj yij rij (9)
dtR P
for this reaction including 8 rate expressions for H2S decomposition j i
and a choice for occurrence of this reaction. With a same procedure,
The above ordinary differential equations were solved by Run-
the total choices of 4 is obtained for the COS recombination reac-
geeKutta order 4 and the numerical algorithm was written in
tion. This procedure was repeated for other reactions in Table 1 but
MATLAB programming software.
for combustion reactions, absence of reactions has not been
The investigation of model results obtained from different re-
considered in calculations as these combustion reactions certainly
action sets indicated that the reaction set of Table 2 was the best
occur in the RF. Therefore total number of possible reaction
possible reaction scheme that could be employed in modeling the
schemes is calculated from multiplication of all reaction alterna-
reaction furnace effluents. It must be noticed that equilibrium
tives as follows:
constants in Table 2 were calculated using 7 coefficient NASA
polynomials whose coefficients were obtained from NASA data
Total number of reaction Schemes ¼ 9 4 6 … 1 (4)
bank (McBride et al., 1993).
The reaction furnace was modeled using different reaction Table 3 compares the calculated reaction furnace effluents with
schemes assuming adiabatic plug flow reactor. Considering adia- the measured experimental data. The mean absolute percentage
batic plug flow assumption and negligible pressure drop, the error (MAPE) of each compound and temperature is also given in
Table 1
Candidate list of reactions in the reaction furnace
Reaction Reference
H2S 4 H2 þ S2 (Adesina et al., 1995; Chin, 2000; Dowling and Clark, 1999; Dowling et al., 1990; Harvey et al., 1998; Hawboldt et al., 2000; Kaloidas and
Papayannakos, 1989; Karan et al., 1999c)
CO þ 0.5S2 4 COS (Karan et al., 1998; Karan et al., 1999a; Klemm and Davis, 1974)
CO2 þ H2 4 CO þ H2O (Bustamante et al., 2005; Graven and Long, 1954; Jones and Lindstedt, 1988; Karan, 1998; Tingey, 1966)
CO þ H2S 4 COS þ H2 (Karan et al., 1999a)
H2S þ 0.5SO2 4 H2O þ 0.75S2 (Monnery et al., 2000; Tesner et al., 1990)
CO2 4 CO þ 0.5O2 (Dryer and Glassman, 1973; Williams et al., 1969)
H2O 4 H2 þ 0.5O2 (Jones and Lindstedt, 1988; Peters, 1979)
CH4 þ 1.5O2 / CO þ 2H2O (Dryer and Glassman, 1973; Westbrook and Dryer, 1981; Williams et al., 1969)
CH4 þ 0.5O2 / CO þ 2H2 (Jones and Lindstedt, 1988; Westbrook and Dryer, 1981)
CH4 þ H2O / CO þ 3H2 (Jones and Lindstedt, 1988)
H2S þ 1.5O2 / SO2 þ H2O (Monnery et al., 2000)
NH3 / 0.5N2 þ 1.5H2 (Monnery et al., 2001)
NH3 þ 0.75O2 / 0.5N2 þ 1.5H2O (Monnery et al., 2001)
C2H4 þ 2O2 / 2CO þ 2H2O (Westbrook and Dryer, 1981)
C2H6 þ 2.5O2 / 2CO þ 3H2O (Westbrook and Dryer, 1981)
C3H8 þ 3.5O2 / 3CO þ 4H2O (Westbrook and Dryer, 1981)
C4H10 þ 4.5O2 / 4CO þ 5H2O (Westbrook and Dryer, 1981)
750 S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757
Table 2
Reaction rate expressions of the chosen kinetic scheme
!
CO2 þ H2 4 CO þ H2O CCO CH2 O (Karan, 1998)
kmol kJ
rCO2 m3 sec
¼ 8:59 1010 exp 270:5
RT mol
CCO2 CH0:5
2
K1eq CH0:5
2
C2H4 þ 2O2 / 2CO þ 2H2O (Westbrook and Dryer, 1981)
rC2 H4 cmmol
3 sec ¼ 2:4 1012 exp RT 30 kcal C 0:1 C 1:65
mol C2 H4 O2
C2H6 þ 2.5O2 / 2CO þ 3H2O (Westbrook and Dryer, 1981)
rC2 H6 cm3 sec ¼ 1:3 10 exp RT mol CC0:1
mol 12 30 kcal
2 H6
CO1:65
2
C3H8 þ 3.5O2 / 3CO þ 4H2O (Westbrook and Dryer, 1981)
rC3 H8 cmmol
3 sec ¼ 1 1012 exp RT 30 kcal C 0:1 C 1:65
mol C3 H8 O2
C4H10 þ 4.5O2 / 4CO þ 5H2O (Westbrook and Dryer, 1981)
rC4 H10 cmmol
3 sec ¼ 8:8 10 11 exp 30 kcal C 0:1 C 1:65
RT mol C4 H10 O2
Table 3
A comparison between the calculated and experimental data at the reaction furnace outlet
Test1 Experimental 10.04 132.59 2.74 6.43 10.31 0.47 5.67 95.30 33.86 1.52 1594.15
Calculated 9.91 134.41 2.79 7.61 12.48 0.12 5.80 93.79 33.11 1.55 1730.27
Test2 Experimental 7.94 144.37 2.07 5.88 9.66 0.29 4.30 82.78 28.97 1.68 1497.15
Calculated 7.40 149.44 1.92 6.85 11.07 0.11 5.34 81.94 27.79 1.74 1639.81
Test3 Experimental 3.86 87.62 1.32 4.80 6.37 0.24 3.28 45.60 20.96 1.05 1481.15
Calculated 3.43 85.99 1.18 5.28 8.04 0.12 2.81 44.56 20.48 1.03 1590.80
Test4 Experimental 5.78 91.42 2.07 6.34 9.82 0.42 4.11 61.47 28.91 1.10 1558.15
Calculated 5.68 94.87 2.12 7.00 10.04 0.14 4.11 61.90 29.04 1.13 1689.60
Test5 Experimental 11.18 111.75 3.52 6.60 11.52 0.60 6.17 121.40 40.02 1.26 1761.15
Calculated 14.84 113.88 4.07 8.00 13.34 0.10 7.47 117.52 38.79 1.28 1874.85
Mean Absolute 10.73 2.48 7.50 15.28 15.95 67.28 12.35 1.76 2.43 2.32 8.07
Percentage Error
this table. As can be seen, except for carbonyl sulfide which has a
significant error, the model has reasonable precision in predicting
the molar flow of other compounds. To improve the model and kf C C
rCO ¼ 0 / ¼ COS total (11)
reduce its error, different reactor networks were analyzed and ki- kb CCO CS2
netic parameter fitting methods were made use of.
On the other hand for COS recombination equilibrium reaction,
A minute review of Table 2 shows that the reaction rate ex-
the equilibrium constant is given by:
pressions related to hydrogen sulfide and carbonyl sulfide decom-
position are not thermodynamically consistent. For equilibrium
reactions, the ratio of forward to backward rate constant (kf/kb) CCOS
equals to equilibrium constant which should be only a function of K¼ (12)
CCO CS0:5
temperature. In thermodynamically inconsistence rate expression, 2
this ratio is a function of concentration or partial pressure of spe- With replacing Equation (11) in Equation (12), we have:
cific component or total concentration and pressure of the system
(Hawboldt et al., 2000).
Based on Table 2, the rate expression of COS recombination is as 0:5
kf CS2
follows: K¼ (13)
kb Ctotal
As it can be seen, the equilibrium constant is a function of
rCO ¼ kf CCO CS2 kb CCOS Ctotal (10) concentration and it is impossible for a consistent reaction.
Therefore the mentioned rate expressions were replaced with their
Where kf and kb are forward and backward rate constants, thermodynamically consistent rate expression forms from Dowling
respectively. Under equilibrium condition: and Clark (1999) and Karan et al. (1999a) as follows:
S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757 751
Table 4
Reaction furnace output using thermodynamically consistent rate expressions in comparison to the measured plant data
Test1 Experimental 10.04 132.59 2.74 6.43 10.31 0.47 5.67 95.30 33.86 1.52 1594.15
Calculated 9.91 134.41 2.79 7.61 12.48 0.12 5.80 93.79 33.11 1.55 1730.27
Test2 Experimental 7.94 144.37 2.07 5.88 9.66 0.29 4.30 82.78 28.97 1.68 1497.15
Calculated 7.40 149.44 1.92 6.85 11.07 0.11 5.34 81.94 27.79 1.74 1639.81
Test3 Experimental 3.86 87.62 1.32 4.80 6.37 0.24 3.28 45.60 20.96 1.05 1481.15
Calculated 3.43 85.99 1.18 5.28 8.04 0.12 2.81 44.56 20.48 1.03 1590.80
Test4 Experimental 5.78 91.42 2.07 6.34 9.82 0.42 4.11 61.47 28.91 1.10 1558.15
Calculated 5.68 94.87 2.12 7.00 10.04 0.14 4.11 61.90 29.04 1.13 1689.60
Test5 Experimental 11.18 111.75 3.52 6.60 11.52 0.60 6.17 121.40 40.02 1.26 1761.15
Calculated 14.84 113.88 4.07 8.00 13.34 0.10 7.47 117.52 38.79 1.28 1874.85
Mean Absolute 56.30 2.46 35.73 3.98 6.96 84.81 42.72 4.79 2.71 2.32 5.77
Percentage Error
752 S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757
Table 5
Results of applying different reactor networks
PFR 56.30 2.46 35.73 3.98 6.96 84.81 42.72 4.79 2.71 2.32 24.28 5.77
CSTR 77.76 9.48 71.83 26.08 14.80 81.09 37.24 2.12 5.68 9.67 33.58 5.67
CSTR þ PFR 56.75 2.50 36.19 3.98 6.97 84.82 42.89 4.75 2.66 6.50 24.39 5.83
CSTR þ CSTR 77.60 9.43 71.48 25.87 14.67 81.14 37.45 2.06 5.63 9.63 33.49 5.67
Table 6
The adjusted kinetic parameters
k E (kJ/mol) k E (kJ/mol)
Table 7
The Calculated reaction furnace effluents after kinetic parameters optimization in comparison to the measured plant data
Test1 Experimental 10.04 132.59 2.74 6.43 10.31 0.47 5.67 95.3 33.86 1.52 1594.15
Calculated 9.62 134.41 2.62 7.51 10.98 0.45 4.88 95.65 34.15 1.55 1731.28
Test2 Experimental 7.94 144.37 2.07 5.88 9.66 0.29 4.30 82.78 28.97 1.68 1497.15
Calculated 6.68 149.45 1.71 6.89 9.89 0.34 4.30 83.92 28.78 1.74 1644.44
Test3 Experimental 3.86 87.62 1.32 4.80 6.37 0.24 3.28 45.60 20.96 1.05 1481.15
Calculated 3.34 85.99 1.11 5.19 7.08 0.29 2.32 45.62 21.11 1.03 1590.04
Test4 Experimental 5.78 91.42 2.07 6.34 9.82 0.42 4.11 61.47 28.91 1.10 1558.15
Calculated 5.81 94.87 2.06 6.79 8.67 0.43 3.59 63.15 29.85 1.13 1686.17
Test5 Experimental 11.18 111.75 3.52 6.60 11.52 0.60 6.17 121.40 40.02 1.26 1761.15
Calculated 15.28 113.88 3.99 7.76 11.44 0.52 6.74 119.11 39.90 1.28 1871.09
Mean Absolute 14.13 2.48 10.45 13.35 6.48 11.56 12.99 1.28 1.16 2.32 8.05
Percentage Error
Table 8
The Calculated waste heat boiler effluents after kinetic parameters optimization in comparison to measured plant data
Test1 Experimental 5.52 132.83 1.95 6.40 13.97 1.27 5.72 96.22 31.62
Calculated 5.84 132.59 2.05 6.43 14.51 1.16 5.67 95.30 31.42
Test2 Experimental 3.87 147.81 1.30 6.04 11.40 0.91 4.45 85.23 27.90
Calculated 6.00 144.37 1.62 5.88 11.60 0.74 4.30 82.78 27.77
Test3 Experimental 2.85 84.78 0.90 4.79 6.66 0.67 2.45 46.27 21.01
Calculated 2.87 87.62 0.99 4.80 7.36 0.57 3.28 45.60 20.30
Test4 Experimental 3.69 93.58 1.37 6.34 11.12 1.12 4.59 62.39 27.69
Calculated 3.69 91.42 1.48 6.34 11.91 1.01 4.11 61.47 27.57
Test5 Experimental 6.31 112.47 2.49 6.59 14.39 1.66 5.71 123.45 38.31
Calculated 5.22 111.75 2.44 6.60 17.48 1.68 6.17 121.4 36.50
Mean Absolute percentage 15.73 1.76 9.88 0.7 8.96 10.57 11.33 1.68 1.93
Error
4. Results and discussions 4.1. Increasing NH3 and O2 molar flow rates
After modeling the reaction furnace and waste heat boiler as the The effects of increasing NH3 and O2 molar flow rates in reaction
front end of Claus unit, the influence of different operating condi- furnace feed were investigated on reaction furnace temperature, H2
tions were studied on reaction furnace outlet temperature, and COS molar flow rate, H2S to SO2 ratio and sulfur conversion
hydrogen and carbonyl sulfide production, sulfur conversion effi- efficiency percentage. Accordingly, the inlet NH3 and O2 molar flow
ciency percentage and H2S to SO2 flow ratio. The NH3 and O2 molar rates varied from 0 to 20 mol/sec for base test 3 which is only made
flow rates, and air to acid gas ratio, were some of these operating of acid gas and air. To put it in another way, sour water gas stripper
conditions that were later allocated much attention in our research and oxygen enrichment flows were not considered in this study.
owing to the following optimization work. Fig. 1 shows the effect of increasing O2 and NH3 molar flow rate
on reaction furnace temperature. As seen, increasing the O2 flow
rate increases the furnace temperature which is favorable for
S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757 753
Table 9
The Calculated waste heat boiler effluents after kinetic parameters optimization in comparison to measured plant data using series modeling
Test1 Experimental 5.52 132.83 1.95 6.40 13.97 1.27 5.72 96.22 31.62
Calculated 5.45 134.41 1.96 7.51 15.15 1.11 4.88 95.65 31.73
Test2 Experimental 3.87 147.81 1.30 6.04 11.40 0.91 4.45 85.23 27.90
Calculated 4.74 149.45 1.37 6.89 11.84 0.68 4.30 83.92 27.64
Test3 Experimental 2.85 84.78 0.90 4.79 6.66 0.67 2.45 46.27 21.01
Calculated 2.40 85.99 0.84 5.19 8.02 0.55 2.32 45.62 20.51
Test4 Experimental 3.69 93.58 1.37 6.34 11.12 1.12 4.59 62.39 27.69
Calculated 3.31 94.87 1.46 6.79 11.17 1.02 3.59 63.15 28.30
Test5 Experimental 6.31 112.47 2.49 6.59 14.39 1.66 5.71 123.45 38.31
Calculated 6.34 113.88 2.75 7.76 20.38 1.77 6.74 119.11 34.80
Mean Absolute Percentage 10.05 1.27 5.89 12.92 14.94 14.19 12.59 1.65 3.01
Error
Fig. 1. Effect of increasing NH3 and O2 molar flow rate on reaction furnace
temperature. Fig. 2. Effect of increasing NH3 and O2 molar flow rate on H2 production at reaction
furnace and waste heat boiler effluents.
Fig. 3. Effect of NH3 and O2 inlet molar flow rates on Sulfur conversion efficiency in 4.2. Air to acid gas flow ratio
reaction furnace and waste heat boiler outlet.
Fig. 7 shows the effect of air to acid gas ratio on the reaction
furnace temperature. The air to acid gas flow ratio ranged from 0.4
to 3, but the summation of these two flows was assumed constant.
According to Fig. 7, increasing the air to acid gas ratio results in the
increment of reaction furnace outlet temperature and the order of
arrangement depends on feed composition including H2S, NH3 and
O2 contents. Based on previous statements about influences of
increasing O2 and NH3 molar flow rate, higher temperatures are
expected for feeds including oxygen enrichment flow, and for feeds
including sour water gas stripper, lower temperatures are expected.
In this study, test 2 with sour water gas stripper was lower than
base test 3, and test 4 including oxygen enrichment flow had higher
reaction furnace temperature in comparison to tests 2 and 3. To
complete hydrogen sulfide combustion, increasing the air to acid
gas ratio outfits is indispensable. The combustion produces high
Fig. 4. Effect of NH3 molar flow rate on H2S and SO2 molar flow rates at reaction
furnace outlet.
Fig. 8. Effect of air to acid gas flow ratio on sulfur conversion efficiency at reaction
Fig. 6. Effect of NH3 and O2 molar flow rates on COS molar flow.
furnace outlet.
5. Optimization
The air to acid gas flow ratio, initial feed temperature, and steam
pressure in the WHB are operating variables that can be changed in
operating Claus furnace in order to maintain H2S/SO2 ratio at 2,
minimize COS emissions and maximize sulfur conversion efficiency
at the waste heat boiler output; such variables can be applied in
maximizing reaction furnace temperature as well. Their variation
ranges are determined in accordance with 0.5e3 for the air to acid
gas flow ratio, 75e375 C for the reaction furnace feed temperature
and 5e30 bar for the steam pressure of waste heat boiler. In present
work, one optimization study was performed where objective
functions including COS emission, H2S to SO2 ratio, reaction furnace
temperature and sulfur production were optimized using reaction
furnace feed temperature, air to acid gas ratio and steam pressure.
Tables 10 and 11 report the results of optimization for all tests, i.e.
the values of decision variables and objective function at optimized
condition. Application of new optimized parameters to the series
model showed that the sulfur production did not increase and
these new conditions had no significant effect on its production
from the thermal stage; but instead the H2S to SO2 ratio, and re-
action furnace temperature access to their optimum values. Opti-
Fig. 7. Effect of air to acid gas flow ratio on reaction furnace outlet temperature.
mum conditions do not change carbonyl sulfide emissions from the
waste heat boiler except for tests 1 and 5 with oxygen enrichment
temperature favorable for decomposition reaction of hydrogen and sour water gas stripper streams where improvements in COS
sulfide and carbonyl sulfide; hydrogen sulfide concentration, emissions were observed.
however, decreases at the same time, resulting in lower decom-
position reaction and sulfur production due to lower decomposi-
tion reaction. Fig. 8 shows sulfur conversion efficiency percentage
as a function of the air to acid gas ratio where it primarily increases
with the air to acid gas ratio, reaches its maximum, then assumes
decreasing trend. Place where maximum is observed for feed Table 10
Optimum parameters for Ultramar reaction furnace
including oxygen enrichment stream happens at lower air to acid
gas flow ratio while for feed including sour water gas stripper oc- Air to Acid Gas Ratio Feed Temperature (K) Steam Pressure (bar)
curs at higher air to acid gas flow ratio. Test1 2.30 470.65 10.6
Test2 2.96 579.75 13.4
Test3 2.02 633.86 16.3
Test4 1.60 565.86 20.1
Test5 1.70 411.40 8.5
756 S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757
Table 11
Values of objective function at optimum condition.
abs (H2S/SO2-2) 1/TRF (1/K) COS(mol/sec) 1/S2 (sec/mol) abs (H2S/SO2-2) 1/TRF (1/K) COS(mol/sec) 1/S2 (sec/mol)
6. Conclusions Chin, H.S.F., 2000. The Fate of Methane in the Reaction Furnace of a Modified Claus
Plant (M.Sc. thesis). department of chemical and petroleum engineering, Cal-
gary, Alberta.
In this work, a suitable reaction scheme was found to model the Chin, H.S.F., Karan, K., Mehrotra, A.K., Behie, L.A., 2001. The fate of methane in a
reaction furnace and waste heat boiler of Claus process. In this claus plant reaction furnace. Can. J. Chem. Eng. 79, 482e490.
modeling, we employed different reactor networks including CSTR Dowling, N.I., Hyne, J.B., Brown, D.M., 1990. Kinetics of the reaction between
hydrogen and sulfur under high-temperature Claus furnace conditions. Ind.
and PFR with optimized kinetic parameters of selected reaction Eng. Chem. Res. 29, 2327e2332.
scheme which reduced the mean percentage error to 7.62%. Fitting Dowling, N.I., Hyrne, J.B., Moon, D.L., Clark, P.O., 1995. An H2S thermal cracking claus
procedure was regulated in a way that the results of kinetic model furnace modification for the production of hydrogen. In: Annual Technical
Meeting.
match with the experimental reaction furnace and waste heat Dowling, N.I., Clark, P.D., 1999. Kinetic modeling of the reaction between hydrogen
boiler effluents. In other words, the developed model can predict and sulfur and opposing H2S decomposition at high temperatures. Ind. Eng.
both reaction furnace and waste heat boiler effluents. So as to Chem. Res. 38, 1369e1375.
Dryer, F.L., Glassman, I., 1973. High-temperature oxidation of CO and CH4. Symp. Int.
survey the effect of oxygen enrichment and sour water gas stripper Combust. 14, 987e1003.
(containing ammonia) on the performance of Claus reaction Graven, W.M., Long, F.J., 1954. Kinetics and mechanisms of the two opposing re-
furnace, different case studies were performed through changing actions of the equilibrium CO þ H2O ¼ CO2 þ H21. J. Am. Chem. Soc. 76,
2602e2607.
the oxygen and ammonia molar flow rates. Results indicated that Harvey, W.S., Davidson, J.H., Fletcher, E.A., 1998. Thermolysis of hydrogen sulfide in
higher levels of oxygen resulted in higher reaction furnace tem- the temperature range 13501600 K. Ind. Eng. Chem. Res. 37, 2323e2332.
peratures, while the presence of ammonia reduced the reaction Hawboldt, K.A., Monnery, W.D., Svrcek, W.Y., 2000. New experimental data and
kinetic rate expression for H2S pyrolysis and re-association. Chem. Eng. Sci. 55,
furnace temperature. The influence of ammonia and oxygen in-
957e966.
crements on sulfur conversion is similar, causing sulfur conversion Ibrahim, S., Al Shoaibi, A., Gupta, A.K., 2013. Role of toluene in hydrogen sulfide
reduction. An optimization study was conducted to investigate the combustion under Claus condition. Appl. Energy 112, 60e66.
optimum operating condition of the reaction furnace and waste Ibrahim, S., Al Shoaibi, A., Gupta, A.K., 2014. Toluene destruction in thermal stage of
Claus reactor with oxygen enriched air. Appl. Energy 115, 1e8.
heat boiler including reaction furnace operating temperature, Javanmardi, N., Fatemi, S., 2015. Kinetic modelling of a commercial sulfur recovery
maximizing sulfur production, minimizing COS emissions and unit based on Claus straight through process: comparison with equilibrium
maintaining H2S/SO2 ratio at constant 2 from the waste heat boiler. model. J. Ind. Eng. Chem. 30, 50e63.
Jones, D., Bhattacharyya, D., Turton, R., Zitney, S.E., 2012. Rigorous kinetic modeling
In this optimization, reaction furnace feed temperature, steam and optimization study of a modified claus unit for an integrated gasification
pressure and air to acid gas flow ratio parameters were used. combined cycle (IGCC) power plant with CO2 capture. Ind. Eng. Chem. Res. 51,
2362e2375.
Jones, W., Lindstedt, R., 1988. Global reaction schemes for hydrocarbon combustion.
Nomenclatures Combust. Flame 73, 233e249.
Kaloidas, V., Papayannakos, N., 1989. Kinetics of thermal, non-catalytic decompo-
sition of hydrogen sulphide. Chem. Eng. Sci. 44, 2493e2500.
A Cross-sectional area (m2) Karan, K., Mehrotra, A.K., Behie, L.A., 1994. Including radiative heat transfer and
h Enthalpy (J/mol) reaction quenching in modeling a claus plant waste heat boiler. Ind. Eng. Chem.
r rate (mol-sec-m) Res. 33, 2651e2655.
Karan, K., 1998. An Experimental and Modeling Study of Homogeneous Gas Phase
C Concentration (mol/m3)
Reactions Occurring in the Modified Claus Process (Ph. D. thesis). Chemical and
F Molar Flow (mol/sec) Petroleum Engineering.
nC Number of components Karan, K., Mehrotra, A.K., Behie, L.A., 1998. COS-forming reaction between CO and
nR Number of reactions sulfur: a high-temperature intrinsic kinetics study. Ind. Eng. Chem. Res. 37,
4609e4616.
P Pressure (Pa) Karan, K., Mehrotra, A.K., Behie, L.A., 1999a. A high-temperature experimental and
R Gas constant (8.314 J/mol K) modeling study of homogeneous gas-phase COS reactions applied to Claus
tR Actual mean residence time(sec) plants. Chem. Eng. Sci. 54, 2999e3006.
Karan, K., Mehrotra, A.K., Behie, L.A., 1999b. Use of new reaction kinetics for COS
T Temperature (K) formation to achieve reduced sulfur emissions from Claus plants. Can. J. Chem.
nij Stoichiometric coefficient of component j in reaction i Eng. 77, 392e398.
Karan, K., Mehrotra, A.K., Behie, L.A., 1999c. On reaction kinetics for the thermal
decomposition of hydrogen sulfide. AIChE J. 45, 383e389.
Subscripts Khudenko, B.M., Gitman, G.M., Wechsler, T.E.P., 1993. Oxygen based claus process
i Reaction number for recovery of sulfur from gases. J. Environ. Eng. 119, 1233e1251.
Klemm, R.B., Davis, D.D., 1974. Flash photolysis-resonance fluorescence kinetics
j Component number study of the reaction sulfur(3p) þ carbonyl sulfide. J. Phys. Chem. 78,
1137e1140.
Manenti, F., Papasidero, D., Cuoci, A., Frassoldati, A., Faravelli, T., Pierucci, S.,
References
Ranzi, E., Buzzi-Ferraris, G., 2012a. Reactor network analysis of Claus furnace
with detailed kinetics. Comput. Aided Chem. Eng. 30, 1007e1012.
Adesina, A.A., Meeyoo, V., Foulds, G., 1995. Thermolysis of hydrogen sulphide in an Manenti, F., Papasidero, D., Bozzano, G., Pierucci, S., Ranzi, E., Buzzi-Ferraris, G.,
open tubular reactor. Int. J. Hydrogen Energy 20, 777e783. 2013a. Total plant integrated optimization of sulfur recovery and steam gen-
Bennett, H.A., Meisen, A., 1973. Hydrogen sulphide e air equilibria under claus eration for Claus processes using detailed kinetic schemes. Comput. Aided
furnace conditions. Can. J. Chem. Eng. 51, 720e724. Chem. Eng. 32, 811e816.
Bustamante, F., Enick, R.M., Killmeyer, R.P., Howard, B.H., Rothenberger, K.S., Manenti, F., Papasidero, D., Frassoldati, A., Bozzano, G., Pierucci, S., Ranzi, E., 2013b.
Cugini, A.V., Morreale, B.D., Ciocco, M.V., 2005. Uncatalyzed and wall-catalyzed Multi-scale modeling of Claus thermal furnace and waste heat boiler using
forward wateregas shift reaction kinetics. AIChE J. 51, 1440e1454.
S. Zarei et al. / Journal of Natural Gas Science and Engineering 31 (2016) 747e757 757
detailed kinetics. Comput. Chem. Eng. 59, 219e225. approach to robust simulation of claus processes in coking plants. Comput.
Manenti, F., Papasidero, D., Bozzano, G., Ranzi, E., 2014. Model-based optimization Aided Chem. Eng. 24, 521e526.
of sulfur recovery units. Comput. Chem. Eng. 66, 244e251. Selim, H., Gupta, A.K., Sassi, M., 2008. Acid gas composition effects on the optimum
Manenti, G., Papasidero, D., Manenti, F., Bozzano, G., Pierucci, S., 2012b. Design of temperature in claus reactor. In: 6th International Energy Conversion Engi-
SRU thermal reactor and waste heat boiler considering recombination re- neering Conference (IECEC).
actions. Procedia Eng. 42, 414e421. Selim, H., Al Shoaibi, A., Gupta, A.K., 2011. Effect of H2S in methane/air flames on
McBride, B.J., Gordon, S., Reno, M.A., 1993. Coefficients for Calculating Thermody- sulfur chemistry and products speciation. Appl. Energy 88, 2593e2600.
namic and Transport Properties of Individual Species. Selim, H., Al Shoaibi, A., Gupta, A.K., 2012a. Fate of sulfur with H2S injection in
Monnery, W.D., Svrcek, W.Y., Behie, L.A., 1993. Modelling the modified claus process methane/air flames. Appl. Energy 92, 8.
reaction furnace and the implications on plant design and recovery. Can. J. Selim, H., Gupta, A.K., Al Shoaibi, A., 2012b. Effect of CO2 and N2 concentration in
Chem. Eng. 71, 711e724. acid gas stream on H2S combustion. Appl. Energy 98, 53e58.
Monnery, W.D., Hawboldt, K.A., Pollock, A., Svrcek, W.Y., 2000. New experimental Selim, H., Ibrahim, S., Al Shoaibi, A., Gupta, A.K., 2013. Effect of oxygen enrichment
data and kinetic rate expression for the Claus reaction. Chem. Eng. Sci. 55, on acid gas combustion in hydrogen/air flames under claus conditions. Appl.
5141e5148. Energy 109, 119e124.
Monnery, W.D., Hawboldt, K.A., Pollock, A.E., Svrcek, W.Y., 2001. Ammonia pyrolysis Selim, H., Ibrahim, S., Al Shoaibi, A., Gupta, A.K., 2014. Investigation of sulfur
and oxidation in the claus furnace. Ind. Eng. Chem. Res. 40, 144e151. chemistry with acid gas addition in hydrogen/air flames. Appl. Energy 113,
Nasato, L.V., 1993. Modelling of Quench Times in the Waste Heat Boiler of a 1134e1140.
Modified Claus Plant (Ph. D. thesis). Chemical and Petroleum Engineering. Tesner, P., Nemirovskii, M., Motyl, D., 1990. Kinetics of the thermal decomposition of
Nasato, L.V., Karan, K., Mehrotra, A.K., Behie, L.A., 1994. Modeling reaction quench hydrogen sulfide at 600e1200 C. Kinet. Catal. 31, 1081e1083.
times in the waste heat boiler of a Claus plant. Ind. Eng. Chem. Res. 33, 7e13. Tingey, G.L., 1966. Kinetics of the water-gas equilibruim Reaction. I. The reaction of
Peters, N., 1979. Premixed burning in diffusion flamesdthe flame zone model of carbon dioxide with hydrogen. J. Phys. Chem. 70, 1406e1412.
libby and economos. Int. J. Heat Mass Transf. 22, 691e703. Westbrook, C.K., Dryer, F.L., 1981. Simplified reaction mechanisms for the oxidation
Pierucci, S., Ranzi, E., Molinari, L., 2004. Modeling a claus process reaction furnace of hydrocarbon fuels in flames. Combust. Sci. Technol. 27, 31e43.
via a radical kinetic scheme. Comput. Aided Chem. Eng. 18, 463e468. Williams, G.C., Hottel, H.C., Morgan, A.C., 1969. The combustion of methane in a jet-
Sames, J.A., Paskall, H.G., Brown, D.M., Chen, M.S.K., Sulkowski, D., 1991. Field mixed reactor. Symp. Int. Combust. 12, 913e925.
Measurements of Hydrogen Production in an Oxygen-enriched Claus Furnace, ZareNezhad, B., Hosseinpour, N., 2008. Evaluation of different alternatives for
pp. 8e105. increasing the reaction furnace temperature of Claus SRU by chemical equi-
Scho€neberger, J., Arellano-Garcia, H., Thielert, H., Wozny, G., 2007. An efficient librium calculations. Appl. Therm. Eng. 28, 738e744.